Accurate Esophageal Gross Tumor Volume Segmentation in PET/CT Using Two-Stream Chained 3D Deep Network Fusion

Jin, Dakai; Guo, Dazhou; Ho, Tsung-Ying; Harrison, Adam P.; Xiao, Jing; Tseng, Chen‐Kan; Lü, Le

doi:10.1007/978-3-030-32245-8_21

Cited by 59 publications

(45 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The use of CNN (1,2) in radiotherapy has been shown to be a state-of-the-art method for organs and tumor segmentation in several disease sites (3)(4)(5)(6)(7)(8)(9)(10)(11). Various notable developments in the use of deep learning methods for organ segmentation have improved the precision of automatic segmentation (12)(13)(14)(15). However, considerable variability in medical images can cause unpredictable errors, even with the use of the best model.…”

Section: Introductionmentioning

confidence: 99%

CNN-Based Quality Assurance for Automatic Segmentation of Breast Cancer in Radiotherapy

et al. 2020

View full text Add to dashboard Cite

Purpose: More and more automatic segmentation tools are being introduced in routine clinical practice. However, physicians need to spend a considerable amount of time in examining the generated contours slice by slice. This greatly reduces the benefit of the tool's automaticity. In order to overcome this shortcoming, we developed an automatic quality assurance (QA) method for automatic segmentation using convolutional neural networks (CNNs). Materials and Methods: The study cohort comprised 680 patients with early-stage breast cancer who received whole breast radiation. The overall architecture of the automatic QA method for deep learning-based segmentation included the following two main parts: a segmentation CNN model and a QA network that was established based on ResNet-101. The inputs were from computed tomography, segmentation probability maps, and uncertainty maps. Two kinds of Dice similarity coefficient (DSC) outputs were tested. One predicted the DSC quality level of each slice ([0.95, 1] for "good," [0.8, 0.95] for "medium," and [0, 0.8] for "bad" quality), and the other predicted the DSC value of each slice directly. The performances of the method to predict the quality levels were evaluated with quantitative metrics: balanced accuracy, F score, and the area under the receiving operator characteristic curve (AUC). The mean absolute error (MAE) was used to evaluate the DSC value outputs. Results: The proposed methods involved two types of output, both of which achieved promising accuracy in terms of predicting the quality level. For the good, medium, and bad quality level prediction, the balanced accuracy was 0.97, 0.94, and 0.89, respectively; the F score was 0.98, 0.91, and 0.81, respectively; and the AUC was 0.96, 0.93, and 0.88, respectively. For the DSC value prediction, the MAE was 0.06 ± 0.19. The prediction time was approximately 2 s per patient. Conclusions: Our method could predict the segmentation quality automatically. It can provide useful information for physicians regarding further verification and revision of automatic contours. The integration of our method into current automatic segmentation pipelines can improve the efficiency of radiotherapy contouring.

show abstract

Section: Introductionmentioning

confidence: 99%

CNN-Based Quality Assurance for Automatic Segmentation of Breast Cancer in Radiotherapy

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Although the datasets are not comparable, in [28] an average DSC score of 0.76 ± 0.13 was obtained on scans of 110 patients, using 5-fold cross validation. In [23] a DSC score of 0.75 ± 0.04 for four patients as the test set has been reported.…”

Section: Discussionmentioning

confidence: 99%

“…That work in an expensive pre-processing step encodes spatial context by computing the signed distance transform maps (SDMs) of the GTV, lymph nodes (LNs) and organs at risks (OARs) and then feeds the results with the CT image into a 3D CNN. In another work Jin et al [28] proposed a two-stream chained 3D CNN fusion pipeline to segment esophageal GTVs using both CT and PET+CT scans. They evaluated their approach by conducting a 5-fold cross validation on scans of 110 patients.…”

Section: Related Workmentioning

confidence: 99%

Esophageal Tumor Segmentation in CT Images Using a Dilated Dense Attention Unet (DDAUnet)

et al. 2021

View full text Add to dashboard Cite

Manual or automatic delineation of the esophageal tumor in CT images is known to be very challenging. This is due to the low contrast between the tumor and adjacent tissues, the anatomical variation of the esophagus, as well as the occasional presence of foreign bodies (e.g. feeding tubes). Physicians therefore usually exploit additional knowledge such as endoscopic findings, clinical history, additional imaging modalities like PET scans. Achieving his additional information is time-consuming, while the results are error-prone and might lead to non-deterministic results. In this paper we aim to investigate if and to what extent a simplified clinical workflow based on CT alone, allows one to automatically segment the esophageal tumor with sufficient quality. For this purpose, we present a fully automatic end-to-end esophageal tumor segmentation method based on convolutional neural networks (CNNs). The proposed network, called Dilated Dense Attention Unet (DDAUnet), leverages spatial and channel attention gates in each dense block to selectively concentrate on determinant feature maps and regions. Dilated convolutional layers are used to manage GPU memory and increase the network receptive field. We collected a dataset of 792 scans from 288 distinct patients including varying anatomies with air pockets, feeding tubes and proximal tumors. Repeatability and reproducibility studies were conducted for three distinct splits of training and validation sets. The proposed network achieved a DSC value of 0.79 ± 0.20, a mean surface distance of 5.4 ± 20.2mm and 95% Hausdorff distance of 14.7 ± 25.0mm for 287 test scans, demonstrating promising results with a simplified clinical workflow based on CT alone.

show abstract

“…Jin et al (10) integrate the RTCT and PET modalities together into a two-stream chained deep fusion framework, which represents a complete workflow for the target delineation in esophageal cancer radiotherapy and pushes forward the state of automated esophageal GTV and CTV segmentation towards a clinically applicable solution. Using extensive five-fold cross-validation on 110 esophageal cancer patients, they also demonstrate that both the proposed two-stream chained segmentation pipeline that effectively fuses the CT and PET modalities via early and late 3D deep-network-based fusion and the PSNN model can significantly improve the accurate GTV segmentation over the previous state-of-the-art work (11). Yousefi S. et al (12) found that the proposed method, dubbed dilated dense attention Unet (DDAUnet), could segment the gross tumor volume with a mean surface distance of 5.4 ± 20.2mm, demonstrating that a simplified clinical workflow based on CT alone could allow to automatically de-lineate the esophageal GTV with acceptable quality.…”

Section: Introductionmentioning

confidence: 87%

Comparison of the Gross Target Volumes Based on Diagnostic PET/CT for Primary Esophageal Cancer

Shi

et al. 2021

Front. Oncol.

View full text Add to dashboard Cite

BackgroundClinically, many esophageal cancer patients who planned for radiation therapy have already undergone diagnostic Positron-emission tomography/computed tomography (PET/CT) imaging, but it remains unclear whether these imaging results can be used to delineate the gross target volume (GTV) of the primary tumor for thoracic esophageal cancer (EC).MethodsSeventy-two patients diagnosed with thoracic EC had undergone prior PET/CT for diagnosis and three-dimensional CT (3DCT) for simulation. The GTV3D was contoured on the 3DCT image without referencing the PET/CT image. The GTVPET-ref was contoured on the 3DCT image referencing the PET/CT image. The GTVPET-reg was contoured on the deformed registration image derived from 3DCT and PET/CT. Differences in the position, volume, length, conformity index (CI), and degree of inclusion (DI) among the target volumes were determined.ResultsThe centroid distance in the three directions between two different GTVs showed no significant difference (P > 0.05). No significant difference was found among the groups in the tumor volume (P > 0.05). The median DI values of the GTVPET-reg and GTVPET-ref in the GTV3D were 0.82 and 0.86, respectively (P = 0.006). The median CI values of the GTV3D in the GTVPET-reg and GTVPET-ref were 0.68 and 0.72, respectively (P = 0.006).ConclusionsPET/CT can be used to optimize the definition of the target volume in EC. However, no significant difference was found between the GTVs delineated based on visual referencing or deformable registration whether using the volume or position. So, in the absence of planning PET–CT images, it is also feasible to delineate the GTV of primary thoracic EC with reference to the diagnostic PET–CT image.

show abstract

Accurate Esophageal Gross Tumor Volume Segmentation in PET/CT Using Two-Stream Chained 3D Deep Network Fusion

Cited by 59 publications

References 18 publications

CNN-Based Quality Assurance for Automatic Segmentation of Breast Cancer in Radiotherapy

CNN-Based Quality Assurance for Automatic Segmentation of Breast Cancer in Radiotherapy

Esophageal Tumor Segmentation in CT Images Using a Dilated Dense Attention Unet (DDAUnet)

Comparison of the Gross Target Volumes Based on Diagnostic PET/CT for Primary Esophageal Cancer

Contact Info

Product

Resources

About